nokogiri 1.10.10 → 1.12.1

Sign up to get free protection for your applications and to get access to all the features.

Potentially problematic release.


This version of nokogiri might be problematic. Click here for more details.

Files changed (216) hide show
  1. checksums.yaml +4 -4
  2. data/Gemfile +3 -0
  3. data/LICENSE-DEPENDENCIES.md +1173 -884
  4. data/LICENSE.md +1 -1
  5. data/README.md +176 -96
  6. data/dependencies.yml +12 -12
  7. data/ext/nokogiri/depend +38 -358
  8. data/ext/nokogiri/extconf.rb +712 -414
  9. data/ext/nokogiri/gumbo.c +584 -0
  10. data/ext/nokogiri/html4_document.c +166 -0
  11. data/ext/nokogiri/html4_element_description.c +294 -0
  12. data/ext/nokogiri/html4_entity_lookup.c +37 -0
  13. data/ext/nokogiri/html4_sax_parser_context.c +119 -0
  14. data/ext/nokogiri/html4_sax_push_parser.c +95 -0
  15. data/ext/nokogiri/libxml2_backwards_compat.c +121 -0
  16. data/ext/nokogiri/nokogiri.c +228 -91
  17. data/ext/nokogiri/nokogiri.h +188 -89
  18. data/ext/nokogiri/test_global_handlers.c +40 -0
  19. data/ext/nokogiri/xml_attr.c +15 -15
  20. data/ext/nokogiri/xml_attribute_decl.c +18 -18
  21. data/ext/nokogiri/xml_cdata.c +13 -18
  22. data/ext/nokogiri/xml_comment.c +19 -26
  23. data/ext/nokogiri/xml_document.c +267 -195
  24. data/ext/nokogiri/xml_document_fragment.c +13 -15
  25. data/ext/nokogiri/xml_dtd.c +54 -48
  26. data/ext/nokogiri/xml_element_content.c +31 -26
  27. data/ext/nokogiri/xml_element_decl.c +22 -22
  28. data/ext/nokogiri/xml_encoding_handler.c +28 -17
  29. data/ext/nokogiri/xml_entity_decl.c +32 -30
  30. data/ext/nokogiri/xml_entity_reference.c +16 -18
  31. data/ext/nokogiri/xml_namespace.c +58 -49
  32. data/ext/nokogiri/xml_node.c +489 -410
  33. data/ext/nokogiri/xml_node_set.c +174 -162
  34. data/ext/nokogiri/xml_processing_instruction.c +17 -19
  35. data/ext/nokogiri/xml_reader.c +197 -172
  36. data/ext/nokogiri/xml_relax_ng.c +52 -28
  37. data/ext/nokogiri/xml_sax_parser.c +112 -112
  38. data/ext/nokogiri/xml_sax_parser_context.c +105 -86
  39. data/ext/nokogiri/xml_sax_push_parser.c +36 -27
  40. data/ext/nokogiri/xml_schema.c +96 -46
  41. data/ext/nokogiri/xml_syntax_error.c +42 -21
  42. data/ext/nokogiri/xml_text.c +13 -17
  43. data/ext/nokogiri/xml_xpath_context.c +158 -73
  44. data/ext/nokogiri/xslt_stylesheet.c +158 -164
  45. data/gumbo-parser/CHANGES.md +63 -0
  46. data/gumbo-parser/Makefile +101 -0
  47. data/gumbo-parser/THANKS +27 -0
  48. data/gumbo-parser/src/Makefile +34 -0
  49. data/gumbo-parser/src/README.md +41 -0
  50. data/gumbo-parser/src/ascii.c +75 -0
  51. data/gumbo-parser/src/ascii.h +115 -0
  52. data/gumbo-parser/src/attribute.c +42 -0
  53. data/gumbo-parser/src/attribute.h +17 -0
  54. data/gumbo-parser/src/char_ref.c +22225 -0
  55. data/gumbo-parser/src/char_ref.h +29 -0
  56. data/gumbo-parser/src/char_ref.rl +2154 -0
  57. data/gumbo-parser/src/error.c +626 -0
  58. data/gumbo-parser/src/error.h +148 -0
  59. data/gumbo-parser/src/foreign_attrs.c +104 -0
  60. data/gumbo-parser/src/foreign_attrs.gperf +27 -0
  61. data/gumbo-parser/src/gumbo.h +943 -0
  62. data/gumbo-parser/src/insertion_mode.h +33 -0
  63. data/gumbo-parser/src/macros.h +91 -0
  64. data/gumbo-parser/src/parser.c +4886 -0
  65. data/gumbo-parser/src/parser.h +41 -0
  66. data/gumbo-parser/src/replacement.h +33 -0
  67. data/gumbo-parser/src/string_buffer.c +103 -0
  68. data/gumbo-parser/src/string_buffer.h +68 -0
  69. data/gumbo-parser/src/string_piece.c +48 -0
  70. data/gumbo-parser/src/svg_attrs.c +174 -0
  71. data/gumbo-parser/src/svg_attrs.gperf +77 -0
  72. data/gumbo-parser/src/svg_tags.c +137 -0
  73. data/gumbo-parser/src/svg_tags.gperf +55 -0
  74. data/gumbo-parser/src/tag.c +222 -0
  75. data/gumbo-parser/src/tag_lookup.c +382 -0
  76. data/gumbo-parser/src/tag_lookup.gperf +169 -0
  77. data/gumbo-parser/src/tag_lookup.h +13 -0
  78. data/gumbo-parser/src/token_buffer.c +79 -0
  79. data/gumbo-parser/src/token_buffer.h +71 -0
  80. data/gumbo-parser/src/token_type.h +17 -0
  81. data/gumbo-parser/src/tokenizer.c +3463 -0
  82. data/gumbo-parser/src/tokenizer.h +112 -0
  83. data/gumbo-parser/src/tokenizer_states.h +339 -0
  84. data/gumbo-parser/src/utf8.c +245 -0
  85. data/gumbo-parser/src/utf8.h +164 -0
  86. data/gumbo-parser/src/util.c +68 -0
  87. data/gumbo-parser/src/util.h +30 -0
  88. data/gumbo-parser/src/vector.c +111 -0
  89. data/gumbo-parser/src/vector.h +45 -0
  90. data/lib/nokogiri.rb +32 -51
  91. data/lib/nokogiri/css.rb +15 -14
  92. data/lib/nokogiri/css/node.rb +1 -0
  93. data/lib/nokogiri/css/parser.rb +64 -63
  94. data/lib/nokogiri/css/parser.y +3 -3
  95. data/lib/nokogiri/css/parser_extras.rb +39 -36
  96. data/lib/nokogiri/css/syntax_error.rb +2 -1
  97. data/lib/nokogiri/css/tokenizer.rb +1 -0
  98. data/lib/nokogiri/css/xpath_visitor.rb +73 -43
  99. data/lib/nokogiri/decorators/slop.rb +1 -0
  100. data/lib/nokogiri/extension.rb +26 -0
  101. data/lib/nokogiri/gumbo.rb +14 -0
  102. data/lib/nokogiri/html.rb +32 -27
  103. data/lib/nokogiri/html4.rb +40 -0
  104. data/lib/nokogiri/{html → html4}/builder.rb +3 -2
  105. data/lib/nokogiri/{html → html4}/document.rb +17 -30
  106. data/lib/nokogiri/{html → html4}/document_fragment.rb +18 -17
  107. data/lib/nokogiri/{html → html4}/element_description.rb +2 -1
  108. data/lib/nokogiri/{html → html4}/element_description_defaults.rb +2 -1
  109. data/lib/nokogiri/{html → html4}/entity_lookup.rb +2 -1
  110. data/lib/nokogiri/{html → html4}/sax/parser.rb +12 -14
  111. data/lib/nokogiri/html4/sax/parser_context.rb +19 -0
  112. data/lib/nokogiri/{html → html4}/sax/push_parser.rb +6 -5
  113. data/lib/nokogiri/html5.rb +473 -0
  114. data/lib/nokogiri/html5/document.rb +74 -0
  115. data/lib/nokogiri/html5/document_fragment.rb +80 -0
  116. data/lib/nokogiri/html5/node.rb +93 -0
  117. data/lib/nokogiri/jruby/dependencies.rb +20 -0
  118. data/lib/nokogiri/syntax_error.rb +1 -0
  119. data/lib/nokogiri/version.rb +3 -109
  120. data/lib/nokogiri/version/constant.rb +5 -0
  121. data/lib/nokogiri/version/info.rb +215 -0
  122. data/lib/nokogiri/xml.rb +36 -36
  123. data/lib/nokogiri/xml/attr.rb +1 -0
  124. data/lib/nokogiri/xml/attribute_decl.rb +1 -0
  125. data/lib/nokogiri/xml/builder.rb +3 -2
  126. data/lib/nokogiri/xml/cdata.rb +1 -0
  127. data/lib/nokogiri/xml/character_data.rb +1 -0
  128. data/lib/nokogiri/xml/document.rb +92 -41
  129. data/lib/nokogiri/xml/document_fragment.rb +5 -6
  130. data/lib/nokogiri/xml/dtd.rb +1 -0
  131. data/lib/nokogiri/xml/element_content.rb +1 -0
  132. data/lib/nokogiri/xml/element_decl.rb +1 -0
  133. data/lib/nokogiri/xml/entity_decl.rb +1 -0
  134. data/lib/nokogiri/xml/entity_reference.rb +1 -0
  135. data/lib/nokogiri/xml/namespace.rb +1 -0
  136. data/lib/nokogiri/xml/node.rb +629 -293
  137. data/lib/nokogiri/xml/node/save_options.rb +1 -0
  138. data/lib/nokogiri/xml/node_set.rb +1 -0
  139. data/lib/nokogiri/xml/notation.rb +1 -0
  140. data/lib/nokogiri/xml/parse_options.rb +12 -3
  141. data/lib/nokogiri/xml/pp.rb +3 -2
  142. data/lib/nokogiri/xml/pp/character_data.rb +1 -0
  143. data/lib/nokogiri/xml/pp/node.rb +1 -0
  144. data/lib/nokogiri/xml/processing_instruction.rb +1 -0
  145. data/lib/nokogiri/xml/reader.rb +9 -12
  146. data/lib/nokogiri/xml/relax_ng.rb +7 -2
  147. data/lib/nokogiri/xml/sax.rb +5 -4
  148. data/lib/nokogiri/xml/sax/document.rb +25 -30
  149. data/lib/nokogiri/xml/sax/parser.rb +1 -0
  150. data/lib/nokogiri/xml/sax/parser_context.rb +1 -0
  151. data/lib/nokogiri/xml/sax/push_parser.rb +1 -0
  152. data/lib/nokogiri/xml/schema.rb +13 -4
  153. data/lib/nokogiri/xml/searchable.rb +25 -16
  154. data/lib/nokogiri/xml/syntax_error.rb +1 -0
  155. data/lib/nokogiri/xml/text.rb +1 -0
  156. data/lib/nokogiri/xml/xpath.rb +4 -5
  157. data/lib/nokogiri/xml/xpath/syntax_error.rb +2 -1
  158. data/lib/nokogiri/xml/xpath_context.rb +1 -0
  159. data/lib/nokogiri/xslt.rb +17 -16
  160. data/lib/nokogiri/xslt/stylesheet.rb +2 -1
  161. data/lib/xsd/xmlparser/nokogiri.rb +1 -0
  162. data/patches/libxml2/{0002-Remove-script-macro-support.patch → 0001-Remove-script-macro-support.patch} +0 -0
  163. data/patches/libxml2/{0003-Update-entities-to-remove-handling-of-ssi.patch → 0002-Update-entities-to-remove-handling-of-ssi.patch} +0 -0
  164. data/patches/libxml2/{0004-libxml2.la-is-in-top_builddir.patch → 0003-libxml2.la-is-in-top_builddir.patch} +1 -1
  165. data/patches/libxml2/0004-use-glibc-strlen.patch +53 -0
  166. data/patches/libxml2/0005-avoid-isnan-isinf.patch +81 -0
  167. data/patches/libxml2/0006-update-automake-files-for-arm64.patch +2511 -0
  168. data/patches/libxml2/0007-Fix-XPath-recursion-limit.patch +31 -0
  169. data/patches/libxslt/0001-update-automake-files-for-arm64.patch +2511 -0
  170. data/patches/libxslt/0002-Fix-xml2-config-check-in-configure-script.patch +19 -0
  171. data/ports/archives/libxml2-2.9.12.tar.gz +0 -0
  172. metadata +139 -161
  173. data/ext/nokogiri/html_document.c +0 -170
  174. data/ext/nokogiri/html_document.h +0 -10
  175. data/ext/nokogiri/html_element_description.c +0 -279
  176. data/ext/nokogiri/html_element_description.h +0 -10
  177. data/ext/nokogiri/html_entity_lookup.c +0 -32
  178. data/ext/nokogiri/html_entity_lookup.h +0 -8
  179. data/ext/nokogiri/html_sax_parser_context.c +0 -116
  180. data/ext/nokogiri/html_sax_parser_context.h +0 -11
  181. data/ext/nokogiri/html_sax_push_parser.c +0 -87
  182. data/ext/nokogiri/html_sax_push_parser.h +0 -9
  183. data/ext/nokogiri/xml_attr.h +0 -9
  184. data/ext/nokogiri/xml_attribute_decl.h +0 -9
  185. data/ext/nokogiri/xml_cdata.h +0 -9
  186. data/ext/nokogiri/xml_comment.h +0 -9
  187. data/ext/nokogiri/xml_document.h +0 -23
  188. data/ext/nokogiri/xml_document_fragment.h +0 -10
  189. data/ext/nokogiri/xml_dtd.h +0 -10
  190. data/ext/nokogiri/xml_element_content.h +0 -10
  191. data/ext/nokogiri/xml_element_decl.h +0 -9
  192. data/ext/nokogiri/xml_encoding_handler.h +0 -8
  193. data/ext/nokogiri/xml_entity_decl.h +0 -10
  194. data/ext/nokogiri/xml_entity_reference.h +0 -9
  195. data/ext/nokogiri/xml_io.c +0 -61
  196. data/ext/nokogiri/xml_io.h +0 -11
  197. data/ext/nokogiri/xml_libxml2_hacks.c +0 -112
  198. data/ext/nokogiri/xml_libxml2_hacks.h +0 -12
  199. data/ext/nokogiri/xml_namespace.h +0 -14
  200. data/ext/nokogiri/xml_node.h +0 -13
  201. data/ext/nokogiri/xml_node_set.h +0 -12
  202. data/ext/nokogiri/xml_processing_instruction.h +0 -9
  203. data/ext/nokogiri/xml_reader.h +0 -10
  204. data/ext/nokogiri/xml_relax_ng.h +0 -9
  205. data/ext/nokogiri/xml_sax_parser.h +0 -39
  206. data/ext/nokogiri/xml_sax_parser_context.h +0 -10
  207. data/ext/nokogiri/xml_sax_push_parser.h +0 -9
  208. data/ext/nokogiri/xml_schema.h +0 -9
  209. data/ext/nokogiri/xml_syntax_error.h +0 -13
  210. data/ext/nokogiri/xml_text.h +0 -9
  211. data/ext/nokogiri/xml_xpath_context.h +0 -10
  212. data/ext/nokogiri/xslt_stylesheet.h +0 -14
  213. data/lib/nokogiri/html/sax/parser_context.rb +0 -16
  214. data/patches/libxml2/0001-Revert-Do-not-URI-escape-in-server-side-includes.patch +0 -78
  215. data/patches/libxml2/0005-Fix-infinite-loop-in-xmlStringLenDecodeEntities.patch +0 -32
  216. data/ports/archives/libxml2-2.9.10.tar.gz +0 -0
@@ -1,10 +0,0 @@
1
- #ifndef NOKOGIRI_HTML_ELEMENT_DESCRIPTION
2
- #define NOKOGIRI_HTML_ELEMENT_DESCRIPTION
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_html_element_description();
7
-
8
- extern VALUE cNokogiriHtmlElementDescription ;
9
-
10
- #endif
@@ -1,32 +0,0 @@
1
- #include <html_entity_lookup.h>
2
-
3
- /*
4
- * call-seq:
5
- * get(key)
6
- *
7
- * Get the HTML::EntityDescription for +key+
8
- */
9
- static VALUE get(VALUE self, VALUE key)
10
- {
11
- const htmlEntityDesc * desc =
12
- htmlEntityLookup((const xmlChar *)StringValueCStr(key));
13
- VALUE klass, args[3];
14
-
15
- if(NULL == desc) return Qnil;
16
- klass = rb_const_get(mNokogiriHtml, rb_intern("EntityDescription"));
17
-
18
- args[0] = INT2NUM((long)desc->value);
19
- args[1] = NOKOGIRI_STR_NEW2(desc->name);
20
- args[2] = NOKOGIRI_STR_NEW2(desc->desc);
21
-
22
- return rb_class_new_instance(3, args, klass);
23
- }
24
-
25
- void init_html_entity_lookup()
26
- {
27
- VALUE nokogiri = rb_define_module("Nokogiri");
28
- VALUE html = rb_define_module_under(nokogiri, "HTML");
29
- VALUE klass = rb_define_class_under(html, "EntityLookup", rb_cObject);
30
-
31
- rb_define_method(klass, "get", get, 1);
32
- }
@@ -1,8 +0,0 @@
1
- #ifndef NOKOGIRI_HTML_ENTITY_LOOKUP
2
- #define NOKOGIRI_HTML_ENTITY_LOOKUP
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_html_entity_lookup();
7
-
8
- #endif
@@ -1,116 +0,0 @@
1
- #include <html_sax_parser_context.h>
2
-
3
- VALUE cNokogiriHtmlSaxParserContext ;
4
-
5
- static void deallocate(xmlParserCtxtPtr ctxt)
6
- {
7
- NOKOGIRI_DEBUG_START(handler);
8
-
9
- ctxt->sax = NULL;
10
-
11
- htmlFreeParserCtxt(ctxt);
12
-
13
- NOKOGIRI_DEBUG_END(handler);
14
- }
15
-
16
- static VALUE
17
- parse_memory(VALUE klass, VALUE data, VALUE encoding)
18
- {
19
- htmlParserCtxtPtr ctxt;
20
-
21
- if (NIL_P(data))
22
- rb_raise(rb_eArgError, "data cannot be nil");
23
- if (!(int)RSTRING_LEN(data))
24
- rb_raise(rb_eRuntimeError, "data cannot be empty");
25
-
26
- ctxt = htmlCreateMemoryParserCtxt(StringValuePtr(data),
27
- (int)RSTRING_LEN(data));
28
- if (ctxt->sax) {
29
- xmlFree(ctxt->sax);
30
- ctxt->sax = NULL;
31
- }
32
-
33
- if (RTEST(encoding)) {
34
- xmlCharEncodingHandlerPtr enc = xmlFindCharEncodingHandler(StringValueCStr(encoding));
35
- if (enc != NULL) {
36
- xmlSwitchToEncoding(ctxt, enc);
37
- if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) {
38
- rb_raise(rb_eRuntimeError, "Unsupported encoding %s",
39
- StringValueCStr(encoding));
40
- }
41
- }
42
- }
43
-
44
- return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
45
- }
46
-
47
- static VALUE parse_file(VALUE klass, VALUE filename, VALUE encoding)
48
- {
49
- htmlParserCtxtPtr ctxt = htmlCreateFileParserCtxt(
50
- StringValueCStr(filename),
51
- StringValueCStr(encoding)
52
- );
53
- return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
54
- }
55
-
56
- static VALUE
57
- parse_doc(VALUE ctxt_val)
58
- {
59
- htmlParserCtxtPtr ctxt = (htmlParserCtxtPtr)ctxt_val;
60
- htmlParseDocument(ctxt);
61
- return Qnil;
62
- }
63
-
64
- static VALUE
65
- parse_doc_finalize(VALUE ctxt_val)
66
- {
67
- htmlParserCtxtPtr ctxt = (htmlParserCtxtPtr)ctxt_val;
68
-
69
- if (ctxt->myDoc)
70
- xmlFreeDoc(ctxt->myDoc);
71
-
72
- NOKOGIRI_SAX_TUPLE_DESTROY(ctxt->userData);
73
- return Qnil;
74
- }
75
-
76
- static VALUE
77
- parse_with(VALUE self, VALUE sax_handler)
78
- {
79
- htmlParserCtxtPtr ctxt;
80
- htmlSAXHandlerPtr sax;
81
-
82
- if (!rb_obj_is_kind_of(sax_handler, cNokogiriXmlSaxParser))
83
- rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::SAX::Parser");
84
-
85
- Data_Get_Struct(self, htmlParserCtxt, ctxt);
86
- Data_Get_Struct(sax_handler, htmlSAXHandler, sax);
87
-
88
- /* Free the sax handler since we'll assign our own */
89
- if (ctxt->sax && ctxt->sax != (xmlSAXHandlerPtr)&xmlDefaultSAXHandler)
90
- xmlFree(ctxt->sax);
91
-
92
- ctxt->sax = sax;
93
- ctxt->userData = (void *)NOKOGIRI_SAX_TUPLE_NEW(ctxt, sax_handler);
94
-
95
- rb_ensure(parse_doc, (VALUE)ctxt, parse_doc_finalize, (VALUE)ctxt);
96
-
97
- return self;
98
- }
99
-
100
- void init_html_sax_parser_context()
101
- {
102
- VALUE nokogiri = rb_define_module("Nokogiri");
103
- VALUE xml = rb_define_module_under(nokogiri, "XML");
104
- VALUE html = rb_define_module_under(nokogiri, "HTML");
105
- VALUE sax = rb_define_module_under(xml, "SAX");
106
- VALUE hsax = rb_define_module_under(html, "SAX");
107
- VALUE pc = rb_define_class_under(sax, "ParserContext", rb_cObject);
108
- VALUE klass = rb_define_class_under(hsax, "ParserContext", pc);
109
-
110
- cNokogiriHtmlSaxParserContext = klass;
111
-
112
- rb_define_singleton_method(klass, "memory", parse_memory, 2);
113
- rb_define_singleton_method(klass, "file", parse_file, 2);
114
-
115
- rb_define_method(klass, "parse_with", parse_with, 1);
116
- }
@@ -1,11 +0,0 @@
1
- #ifndef NOKOGIRI_HTML_SAX_PARSER_CONTEXT
2
- #define NOKOGIRI_HTML_SAX_PARSER_CONTEXT
3
-
4
- #include <nokogiri.h>
5
-
6
- extern VALUE cNokogiriHtmlSaxParserContext;
7
-
8
- void init_html_sax_parser_context();
9
-
10
- #endif
11
-
@@ -1,87 +0,0 @@
1
- #include <html_sax_push_parser.h>
2
-
3
- /*
4
- * call-seq:
5
- * native_write(chunk, last_chunk)
6
- *
7
- * Write +chunk+ to PushParser. +last_chunk+ triggers the end_document handle
8
- */
9
- static VALUE native_write(VALUE self, VALUE _chunk, VALUE _last_chunk)
10
- {
11
- xmlParserCtxtPtr ctx;
12
- const char * chunk = NULL;
13
- int size = 0;
14
-
15
-
16
- Data_Get_Struct(self, xmlParserCtxt, ctx);
17
-
18
- if(Qnil != _chunk) {
19
- chunk = StringValuePtr(_chunk);
20
- size = (int)RSTRING_LEN(_chunk);
21
- }
22
-
23
- if(htmlParseChunk(ctx, chunk, size, Qtrue == _last_chunk ? 1 : 0)) {
24
- if (!(ctx->options & XML_PARSE_RECOVER)) {
25
- xmlErrorPtr e = xmlCtxtGetLastError(ctx);
26
- Nokogiri_error_raise(NULL, e);
27
- }
28
- }
29
-
30
- return self;
31
- }
32
-
33
- /*
34
- * call-seq:
35
- * initialize_native(xml_sax, filename)
36
- *
37
- * Initialize the push parser with +xml_sax+ using +filename+
38
- */
39
- static VALUE initialize_native(VALUE self, VALUE _xml_sax, VALUE _filename,
40
- VALUE encoding)
41
- {
42
- htmlSAXHandlerPtr sax;
43
- const char * filename = NULL;
44
- htmlParserCtxtPtr ctx;
45
- xmlCharEncoding enc = XML_CHAR_ENCODING_NONE;
46
-
47
- Data_Get_Struct(_xml_sax, xmlSAXHandler, sax);
48
-
49
- if(_filename != Qnil) filename = StringValueCStr(_filename);
50
-
51
- if (!NIL_P(encoding)) {
52
- enc = xmlParseCharEncoding(StringValueCStr(encoding));
53
- if (enc == XML_CHAR_ENCODING_ERROR)
54
- rb_raise(rb_eArgError, "Unsupported Encoding");
55
- }
56
-
57
- ctx = htmlCreatePushParserCtxt(
58
- sax,
59
- NULL,
60
- NULL,
61
- 0,
62
- filename,
63
- enc
64
- );
65
- if(ctx == NULL)
66
- rb_raise(rb_eRuntimeError, "Could not create a parser context");
67
-
68
- ctx->userData = NOKOGIRI_SAX_TUPLE_NEW(ctx, self);
69
-
70
- ctx->sax2 = 1;
71
- DATA_PTR(self) = ctx;
72
- return self;
73
- }
74
-
75
- VALUE cNokogiriHtmlSaxPushParser;
76
- void init_html_sax_push_parser()
77
- {
78
- VALUE nokogiri = rb_define_module("Nokogiri");
79
- VALUE html = rb_define_module_under(nokogiri, "HTML");
80
- VALUE sax = rb_define_module_under(html, "SAX");
81
- VALUE klass = rb_define_class_under(sax, "PushParser", cNokogiriXmlSaxPushParser);
82
-
83
- cNokogiriHtmlSaxPushParser = klass;
84
-
85
- rb_define_private_method(klass, "initialize_native", initialize_native, 3);
86
- rb_define_private_method(klass, "native_write", native_write, 2);
87
- }
@@ -1,9 +0,0 @@
1
- #ifndef NOKOGIRI_HTML_SAX_PUSH_PARSER
2
- #define NOKOGIRI_HTML_SAX_PUSH_PARSER
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_html_sax_push_parser();
7
-
8
- extern VALUE cNokogiriHtmlSaxPushParser ;
9
- #endif
@@ -1,9 +0,0 @@
1
- #ifndef NOKOGIRI_XML_ATTR
2
- #define NOKOGIRI_XML_ATTR
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_attr();
7
-
8
- extern VALUE cNokogiriXmlAttr;
9
- #endif
@@ -1,9 +0,0 @@
1
- #ifndef NOKOGIRI_XML_ATTRIBUTE_DECL
2
- #define NOKOGIRI_XML_ATTRIBUTE_DECL
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_attribute_decl();
7
-
8
- extern VALUE cNokogiriXmlAttributeDecl;
9
- #endif
@@ -1,9 +0,0 @@
1
- #ifndef NOKOGIRI_XML_CDATA
2
- #define NOKOGIRI_XML_CDATA
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_cdata();
7
-
8
- extern VALUE cNokogiriXmlCData;
9
- #endif
@@ -1,9 +0,0 @@
1
- #ifndef NOKOGIRI_XML_COMMENT
2
- #define NOKOGIRI_XML_COMMENT
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_comment();
7
-
8
- extern VALUE cNokogiriXmlComment;
9
- #endif
@@ -1,23 +0,0 @@
1
- #ifndef NOKOGIRI_XML_DOCUMENT
2
- #define NOKOGIRI_XML_DOCUMENT
3
-
4
- #include <nokogiri.h>
5
-
6
- struct _nokogiriTuple {
7
- VALUE doc;
8
- st_table *unlinkedNodes;
9
- VALUE node_cache;
10
- };
11
- typedef struct _nokogiriTuple nokogiriTuple;
12
- typedef nokogiriTuple * nokogiriTuplePtr;
13
-
14
- void init_xml_document();
15
- VALUE Nokogiri_wrap_xml_document(VALUE klass, xmlDocPtr doc);
16
-
17
- #define DOC_RUBY_OBJECT_TEST(x) ((nokogiriTuplePtr)(x->_private))
18
- #define DOC_RUBY_OBJECT(x) (((nokogiriTuplePtr)(x->_private))->doc)
19
- #define DOC_UNLINKED_NODE_HASH(x) (((nokogiriTuplePtr)(x->_private))->unlinkedNodes)
20
- #define DOC_NODE_CACHE(x) (((nokogiriTuplePtr)(x->_private))->node_cache)
21
-
22
- extern VALUE cNokogiriXmlDocument ;
23
- #endif
@@ -1,10 +0,0 @@
1
- #ifndef NOKOGIRI_XML_DOCUMENT_FRAGMENT
2
- #define NOKOGIRI_XML_DOCUMENT_FRAGMENT
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_document_fragment();
7
-
8
- extern VALUE cNokogiriXmlDocumentFragment;
9
- #endif
10
-
@@ -1,10 +0,0 @@
1
- #ifndef NOKOGIRI_XML_DTD
2
- #define NOKOGIRI_XML_DTD
3
-
4
- #include <nokogiri.h>
5
-
6
- extern VALUE cNokogiriXmlDtd;
7
-
8
- void init_xml_dtd();
9
-
10
- #endif
@@ -1,10 +0,0 @@
1
- #ifndef NOKOGIRI_XML_ELEMENT_CONTENT
2
- #define NOKOGIRI_XML_ELEMENT_CONTENT
3
-
4
- #include <nokogiri.h>
5
-
6
-
7
- VALUE Nokogiri_wrap_element_content(VALUE doc, xmlElementContentPtr element);
8
- void init_xml_element_content();
9
-
10
- #endif
@@ -1,9 +0,0 @@
1
- #ifndef NOKOGIRI_XML_ELEMENT_DECL
2
- #define NOKOGIRI_XML_ELEMENT_DECL
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_element_decl();
7
-
8
- extern VALUE cNokogiriXmlElementDecl;
9
- #endif
@@ -1,8 +0,0 @@
1
- #ifndef NOKOGIRI_XML_ENCODING_HANDLER
2
- #define NOKOGIRI_XML_ENCODING_HANDLER
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_encoding_handler();
7
-
8
- #endif
@@ -1,10 +0,0 @@
1
- #ifndef NOKOGIRI_XML_ENTITY_DECL
2
- #define NOKOGIRI_XML_ENTITY_DECL
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_entity_decl();
7
-
8
- extern VALUE cNokogiriXmlEntityDecl;
9
- #endif
10
-
@@ -1,9 +0,0 @@
1
- #ifndef NOKOGIRI_XML_ENTITY_REFERENCE
2
- #define NOKOGIRI_XML_ENTITY_REFERENCE
3
-
4
- #include <nokogiri.h>
5
-
6
- void init_xml_entity_reference();
7
-
8
- extern VALUE cNokogiriXmlEntityReference;
9
- #endif
@@ -1,61 +0,0 @@
1
- #include <xml_io.h>
2
-
3
- static ID id_read, id_write;
4
-
5
- VALUE read_check(VALUE *args) {
6
- return rb_funcall(args[0], id_read, 1, args[1]);
7
- }
8
-
9
- VALUE read_failed(void) {
10
- return Qundef;
11
- }
12
-
13
- int io_read_callback(void * ctx, char * buffer, int len) {
14
- VALUE string, args[2];
15
- size_t str_len, safe_len;
16
-
17
- args[0] = (VALUE)ctx;
18
- args[1] = INT2NUM(len);
19
-
20
- string = rb_rescue(read_check, (VALUE)args, read_failed, 0);
21
-
22
- if (NIL_P(string)) return 0;
23
- if (string == Qundef) return -1;
24
- if (TYPE(string) != T_STRING) return -1;
25
-
26
- str_len = (size_t)RSTRING_LEN(string);
27
- safe_len = str_len > (size_t)len ? (size_t)len : str_len;
28
- memcpy(buffer, StringValuePtr(string), safe_len);
29
-
30
- return (int)safe_len;
31
- }
32
-
33
- VALUE write_check(VALUE *args) {
34
- return rb_funcall(args[0], id_write, 1, args[1]);
35
- }
36
-
37
- VALUE write_failed(void) {
38
- return Qundef;
39
- }
40
-
41
- int io_write_callback(void * ctx, char * buffer, int len) {
42
- VALUE args[2], size;
43
-
44
- args[0] = (VALUE)ctx;
45
- args[1] = rb_str_new(buffer, (long)len);
46
-
47
- size = rb_rescue(write_check, (VALUE)args, write_failed, 0);
48
-
49
- if (size == Qundef) return -1;
50
-
51
- return NUM2INT(size);
52
- }
53
-
54
- int io_close_callback(void * ctx) {
55
- return 0;
56
- }
57
-
58
- void init_nokogiri_io() {
59
- id_read = rb_intern("read");
60
- id_write = rb_intern("write");
61
- }